Ionisation 
73 
66-69] 
temperature is increased sixteen-fold. Or, to put the same thing in the form 
immediately suited to our problem, the temperature corresponding to a given 
pressure is decreased to one-sixteenth. 
The total mass of the sixteen moving parts is equal to the original mass 
of the nitrogen molecule of molecular weight 28. Thus we can regard the 
positive nuclei and negative electrons as forming a gas of mean molecular 
weight or T75. If the original gas had been hydrogen, the mean molecular 
weight of the broken pieces would have been 0*5; similarly the mean molecular 
weight of broken up helium is 1*33, of broken up calcium is 1*90, of broken 
up iron 2*07, and of broken up lead 2'50. There is of course very much greater 
equality between the mean molecular weights of the broken up pieces of 
molecules than between those of the original molecules themselves, since the 
more massive molecules break into a greater number of separate fragments 
than the less massive ones; indeed, the atomic number of an atom of atomic 
weight n in general approximates to \n except for hydrogen and the very 
massive atoms, and in consequence the mean molecular weight of the broken 
mixture can never differ very greatly from 2 . 
For a general mixture of elements such as are known to us on earth, we 
should not be far wrong in assuming a mean molecular weight of 2 for the 
broken up fragments. The corresponding value of R/ntfi is 4T3 x 10 7 in 
place of the much lower value assumed by Emden and used above, and on 
inserting this value in equation ( 66 * 1 ) we obtain ® = 7'76 x 10 6 degrees. The 
central temperature T c is now found to be 
T e — 31*5 million degrees, 
the values of p c and p c remaining unaltered. 
We shall find reasons for supposing that the stars in general do not 
consist of a general mixture of elements of this kind, but that the great 
majority of their atoms have atomic weights comparable with that of uranium 
and the radioactive elements. We shall further find that in all probability the 
atoms are not completely broken up, although very nearly so. When we allow 
for these new factors the mean molecular weights of the broken up fragments 
is about 2*6 rather than 2 , and this increases the value just calculated for T c 
to about 41 million degrees. We shall find that further adjustments have to 
be made in the value of T c , but these nearly neutralise one another leaving > 
the final value of T c in the neighbourhood of 40 million degrees. 
69. There would seem to be little room for doubt that the foregoing view 
of the interior structure of a star is the correct one. Convincing evidence in 
its favour is provided by the high density of many stars, and in particular, as 
Eddington first pointed out, by that of the companion of Sirius, which direct 
observation shews to be something like 50,000. It is inconceivable that such 
high densities could occur in a gas formed of complete atoms or molecules, 
although it is quite natural that they should be found when the molecules of